metadata
license: apache-2.0
base_model: distilbert/distilbert-base-uncased
tags:
- trl
- reward-trainer
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: distilbert_social_reasoning_reward_model
results: []
distilbert_social_reasoning_reward_model
This model is a fine-tuned version of distilbert/distilbert-base-uncased on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.6309
- Accuracy: 0.6958
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0005
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.6618 | 0.24 | 10 | 0.6505 | 0.6725 |
| 0.6357 | 0.48 | 20 | 0.6373 | 0.6497 |
| 0.6457 | 0.72 | 30 | 0.6226 | 0.6725 |
| 0.646 | 0.96 | 40 | 0.6437 | 0.6778 |
| 0.6448 | 1.2 | 50 | 0.7565 | 0.6287 |
| 0.6339 | 1.44 | 60 | 0.6365 | 0.6655 |
| 0.6207 | 1.68 | 70 | 0.6694 | 0.6778 |
| 0.6217 | 1.92 | 80 | 0.6351 | 0.6340 |
| 0.5928 | 2.16 | 90 | 0.7245 | 0.6497 |
| 0.5938 | 2.4 | 100 | 0.6739 | 0.6497 |
| 0.5873 | 2.63 | 110 | 0.6811 | 0.6357 |
| 0.5442 | 2.87 | 120 | 0.6774 | 0.6375 |
Framework versions
- Transformers 4.38.2
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2